equinor / snapwell

The Snapwell wellpath optimization tool

Snapwell is free software: you can redistribute it and/or modify it under
the terms of the GNU General Public License as published by the Free
Software Foundation, either version 3 of the License, or (at your option)
any later version.

Snapwell is distributed in the hope that it will be useful, but WITHOUT ANY
WARRANTY; without even the implied warranty of MERCHANTABILITY or FITNESS
FOR A PARTICULAR PURPOSE.

See the GNU General Public License at <http://www.gnu.org/licenses/gpl.html>
for more details.

Installation and usage

Snappwell is available on pypi and can be installed using pip install snapwell

On snapwell

Snapwell is a project for depth optimizing wellpaths by changing the TVD according to OWC. What exactly defines OWC at a given (x,y)-coordinate can be specified as number r between 0 and 1, and OWC is then the lowest point with SWAT=r. Future implementations might also take the permeability into account.

Changes in TVD are typically done by using a max increase of 0.5 m/30m to avoid making vertical segments.

Should avoid putting the majority of the well path in cells with non producible permeability.

Fileformat

snapwell input file

The filepath is relative to the config file, not the current working directory.

The config file is a yaml-file having the following format:

grid_file: "NORNE_ATW2013.EGRID"
restart_file: "NORNE_ATW2013.UNRST"
init_file: "NORNE_ATW2013.INIT"

output_dir: "norne-mid"
overwrite: False
delta_z: 0.003
log_keywords: ["PERMX", "OWC", "TVD_DIFF"]

wellpath_files:
  - {well_file: "norne-test-1-mid.w", date: "1997-05-1"}
  - {well_file: "norne-test-2-mid.w", date: "1998-01-1"}

• Two keywords are mandatory, grid_file and restart_file, and should be followed by a path to the grid (GRID/EGRID) and restart files (UNRST), respectively (see example below).
• The init_file is required for permeability output, and should be a path to an init file.
• The following additional keywords are supported:
  • output --- the output directory (path), is set to cwd by default
  • overwrite --- a boolean (True/False) specifying whether the program is allowed to overwrite already existing files. Defaulted to False; use with care.
  • delta_z --- the maximum elevation per 100 meters, given as a float. 0.0167 is 0.5m/30m.
  • owc_definition --- The keyword used for saturation and threshold value used to determine owc

owc_definition:
  keyword: "SWAT"
  value: 0.7

• owc_offset --- a number (eg. 0.5). Will place the wellpoint 0.5m above OWC
• log_keywords --- list of values written to the well file:
  • OWC --- actual OWC for given date and (x,y) coordinate
  • OLD_TVD --- the old TVD specified in the input Wellpath
  • DIFF --- The difference between OLD_TVD and new TVD
  • SWAT, SOIL, SGAS, PERMX --- value in wellpoint cell for keyword
  • LENGTH --- the distance between this (x,y) coordinate and the previous
• wellpath_files --- list of files in the wellpath format
  • well_file --- path to the wellpath file
  • date --- YYYY-MM-DD The date to look up values for in the restart file.
  • window_depth --- The depth to start optimizing the well for.
  • depth_type --- Type of depth in the well file, either "MD" or "TVD".

WellPath format (RMS well format)

The wellpath format is given as follows: (blank lines and lines starting with -- ignored)

<version>
<well type>
<well name> <x> <y> <rkb>
<number of well logs>
<log name 1> <unit> <scale>
...
<log name n> <unit> <scale>
<x> <y> <z> <p1> ... <pn>
...
<x> <y> <z> <p1> ... <pn>

An example is the following:

1.0
PRODUCTION
TEST_WELL 1068.000 0.000 0.0000
3
MD 1 lin
Incl 1 lin
Az 1 lin
1068.0    0.0    38.0    38.0    109.0    359.0
1069.0    5.0    39.0    50.0    109.0    359.0

Here, the X, Y, and Z numbers are UTM* easting, UTM northing, UTM TVD (true vertical depth), and then follow the three log parameters defined, MD, INC and AZ, being measured_depth, inclination and azimuth.

We safely ignore all but the three first numbers in each row (actually, only x and y are really interesting).

• UTM is Universal Transverse Mercator coordinate system

Run tests

tox is used as the test facilitator, to run the full test suite:

# Test
pip install tox
tox

or to run it for a particular Python version (in this case Python 3.7):

# Test
pip install tox
tox -e py37

or to run it for a the current Python version on PATH:

# Test
pip install tox
tox -e py

pytest is used as the test runner, so for quicker iteration it is possible to run:

# Test
pytest

this requires that the pytest is installed.

# Install pytest
pip install pytest

pre-commit is used to comply with the formatting standards. The complete formatting tests can be run with:

pip install tox
tox -e style

See .pre-commit-config.yaml for the complete steps.

pre-commit can also provide git hooks to run on every commit to avoid commiting with formatting errors. This will only run on the diff so is quite fast. To configure this, run:

pip install pre-commit
pre-commit install

After this the hook will run on every commit.

If you would like to remove the hooks, run:

pre-commit uninstall

snapwell's Issues

Remove dependency on ConfigParser from libres

The configuration should be replaced with yaml

diff is reset, but never used

This diff is reset, but never used:

snapwell/snapwell/snapecl.py

Line 326 in 8233707

diff = 0

Instead we do logs["TVD_DIFF"].append(new_tvd - z) should that be logs["TVD_DIFF"].append(diff)?

Move tests into repo

Tests already exists, we need to port them, and make sure that they run.

There are no tests using init files

Manual testing with komodoenv

We need to make sure snapwell works in the required context, so need to test it in a komodoenv

Always searching from the bottom up

When finding the contact to snap to we are always searching from the bottom up, this causes a problem when looking for the GOC

snapwell/snapwell/snapecl.py

Line 227 in b500625

threshold_idx = first_swat_below_treshold(col, threshold=threshold)

Replace `snapshottest` with `syrupy`

In case we ever want to start testing on Python 3.12, we'll likely need to drop snapshottest, a Pytest plugin.

It uses imp which is not compatible with Python 3.12. They seem not to be about to fix it: syrusakbary/snapshottest#166

It's easy to switch to syrupy, which has the same API. Migration instructions: https://github.com/tophat/syrupy?tab=readme-ov-file#migration-from-snapshottest

Add test

Add test that

1. Downloads and builds ert
2. Downloads Norne grid and restart file
3. Run an integration test / validation test / smoke test on a single config path

There are no tests using saturation keywords

Keywords "SWAT", "SOIL", and "SGAS" have special behavior which is never tested.

Make ERT hooks

Should be optional, for example installed with pip install snapwell[ert-hooks]

Allow the user to constrain snapping within zones

There is no support in snapwell for confining the well to a zone or region, aside for the fact that it will stay clear of inactive cells. It would be good to support some basic constraints, such as within a zone. This would expand the application to many more use cases.

Add to scout repos

Need to add snapwell to the list of repos maintained by scout

Deploy on pypi

Allow snapwell to complete succesfully even though some snaps fail

Because snapwell is being used as a forward model, a -1 return code might not be wanted when at least some operations finish successfully.

A couple of options as far as I can see:

• Add an option to only fail if a certain number (or percentage) of snaps fail (command line probably)
• We have two different entry points, we could have slight different behavior in the ert entry point.

Setting depthtype without matching column in well file gives uncought error

enterSnapMode assumes that well_path[well_path.depthType()] does not result in a KeyError which it currently can. If this is the case, a good error message should be displayed rather than resulting in a uncaught error which is the current behavior.

Several tests in test_snapconfig.py are failing

Several tests in test_snapconfig.py are failing and are marked as xfail, we need to figure out why. Same error message, so probably the same reason.

Relevant tests:

• snapwell/tests/test_snapconfig.py

  Line 54 in f07c25b

  def test_ParseLOGS(self):

• snapwell/tests/test_snapconfig.py

  Line 76 in f07c25b

  def test_ParseConfFull(self):

• snapwell/tests/test_snapconfig.py

  Line 124 in f07c25b

  def test_ParseSettings(self):

• snapwell/tests/test_snapconfig.py

  Line 160 in f07c25b

  def test_OwcDefinition(self):

Several tests in test_snapwell_norne are failing

Some tests are marked as xfail in test_snapwell_norne.py, we need to figure out why.

Relevant tests:

• snapwell/tests/test_snapwell_norne.py

  Line 22 in f07c25b

  def test_snapwell_invalid_arg_throws(self):

• snapwell/tests/test_snapwell_norne.py

  Line 28 in f07c25b

  def test_snapwell_well_ok(self):

Set up CI pipeline

Must test python 3.6, 3.7 and 3.8

There are no tests using inactive cells

Refactor: ResSlice

Make snapecl construct a ResSlice (wall) of the wellpath before running algorithm.

Instead of running findOwc and activeColumn successively on the wellpoints, construct the entire wall first (with inactive cells).

Consider splitting the wall vertically and adjusting for height (TVD).

This facilitates the use of DP / genetic algorithms.

Let RESTART files be optional in snapwell if INIT file is provided

In snapwell, if RESTART is not provided but INIT is, we may still perform snapping, but using the SWATINIT keyword, and without a given date.

This might be interesting for users without a proper UNRST file.

What needs to be done: Check in snapecl.snap whether argument rest (restart EclFile) is None and init (init EclFile) is not None. If so, we need to ignore (and probably output) all dates, and use the SWATINIT keyword instead of SWAT. Should this be a user concern? I.e., should the user be forced to set SWATINIT as OWC_DEFINITION, should we set SWATINIT as keyword, or should we append INIT to the provided keyword?